Data transmissions in enterprise optical communication systems have not relied on overly-complicated encoding and/or decoding schemes because the technology has been more than sufficient to support desired data transmission rates. However, as computing devices become faster and the need for increased data transmission rates is realized, the physical limits of optical devices will become a limiting factor. Accordingly, optical communication systems will begin heading toward the use of more complicated encoding and decoding schemes.
Pulse-Amplitude Modulation (PAM) is a form of signal modulation where the message information is encoded in the amplitude of a series of signal pulses. It is pulse modulation scheme in which the amplitudes of a train of carrier pulses are varied according to the sample value of the message signal. Demodulation of a PAM-encoded signal is performed by detecting the amplitude level of the carrier at every symbol period.
In a PAM-4-based optical link, two Non-Return-to-Zero (NRZ)-coded two-level signals are combined together in a PAM-4 encoder to create a single PAM four-level signal. The PAM-4 signal is the signal that is ultimately transmitted across a communication network (e.g., through fiber optics). An advantage of a PAM-4 encoding scheme is that the four-level code utilizes the same baud, or symbol rate, of either of the two NRZ codes while containing twice the information of either. This is an attractive solution when the components of the link are baud rate limited, as is often the case for very high-speed fiber links.
Traditional PAM-4 signaling has a strict linearity requirement. Specifically, any PAM signal (e.g., PAM-N, where N is an integer greater than or equal to 4) has been traditionally constrained by the requirement that each signal level is uniformly spaced apart from adjacent signal levels. A conventional PAM-4 encoder translates two NRZ signals (DA and DB) to a PAM-4 signal via a table as shown in FIG. 1A, where a constant value, Ds, is used to define the spacing between adjacent levels. In other words, as shown in FIG. 1B, the base level where DA and DB are both ‘0’ results in a PAM-4 signal of V0. The next level, where DA is ‘0’, but DB is ‘1’ results in a PAM-4 signal of V0+Ds. Each subsequent level is greater than the previous level by the constant, Ds.
A PAM-4 receiver then decodes the PAM-4 signal received from the encoder and recovers the original two NRZ (DA and DB) data streams. The receiver samples the PAM-4 signal at N-1 (e.g., 3 points in a PAM-4 signal) at a common sample time (e.g., ts) and performs an inverse mapping to the encoder table of FIG. 1A. In this way, the PAM-4 encoder behaves much like a Digital-to-Analog converter and the PAM-4 receiver behaves much like an Analog-to-Digital converter.
The strict linearity of encoding and sampling works well for pure electronic systems whose behavior is relatively linear, but the strict linearity presents a number of problems in optical systems due to non-idealities and the non-linear behavior of optical components.